http://www.nspllc.com/New%20Pages/Reliable%20IP%20Nodes.pdf
Argues that by going from a dual-router POP design to a
single redundant router configuration, I can reduce annual
downtime costs by 93% (?) and reduce CAPEX and OPEX (seems
logical).
Also mentions that the single redundant router can get
closer to the ever-elusive 5-9's.
Anyone doing this in their network? Is there validity in the
claims in this white paper? Anyone looked at the Alcatel
product that apparently funded this paper?
I'd believe the numbers in this, what I don't believe yet
though is that there are products available that true give
the same level of redundancy that having two boxes does...
Regards,
Neil.
I've always found that these types of papers and products always miss one
big area of failure, at least in my experience. What happens when the
highly redundant device is skewered by a fork lift? Yes, I've had this
happen. At least if you have a dual router config, and separate those
routers physically, you have a chance of surviving such problems.
Brandon Ross wrote:
I've always found that these types of papers and products always miss one
big area of failure, at least in my experience. What happens when the
highly redundant device is skewered by a fork lift? Yes, I've had this
happen. At least if you have a dual router config, and separate those
routers physically, you have a chance of surviving such problems.
Apparently your skeweree was insufficiently armored -- an obvious
design flaw.
Where do you draw the line? Probability trees aside, you design your
system and take your chances.
Peter E. Fry
Apparently your skeweree was insufficiently armored -- an obvious
design flaw.
LOL, indeed.
Where do you draw the line? Probability trees aside, you design your
system and take your chances.
Of course, you just do the best you can do with the budget that you have
and the business requirements. To date I haven't seen a significant
benefit to using an integrated redundant device, all things considered.